Pith. sign in

REVIEW 2 major objections 6 minor 60 references

This chapter argues that machine learning has reached the point where it can diagnose operational faults in instruments like LIGO and replace slow physics simulations with fast generative models, improving both sensitivity and design speed.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-02 01:32 UTC pith:YS5GROM3

load-bearing objection A competent review chapter with no new results; the operations half is honestly summarized, but the design-side conclusion overreaches beyond what the cited GAN work actually establishes. the 2 major comments →

arxiv 2607.14619 v1 pith:YS5GROM3 submitted 2026-07-16 physics.ins-det

Machine Learning for Complex Instrument Design and Optimization

Barry C. Barish , Jonathan W. Richardson , Evangelos E. Papalexakis , Rutuja Gurav This is my paper
classification physics.ins-det
keywords machine learninginstrument diagnosticsLIGOnoise transientslock losstensor factorizationisolation forestgenerative adversarial networks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The chapter's central claim is that machine learning has moved beyond analyzing an experiment's science data and can now learn about the instrument itself, using the thousands of auxiliary channels that record its state. It makes the case with two end-to-end pipelines: tensor factorization that selects 'witness' channels to veto noise glitches in LIGO data, and an isolation-forest outlier detector that finds precursors to lock-loss control failures. For design, it argues generative adversarial networks can stand in for expensive physics simulators, citing a calorimeter-shower application that matches mean energy patterns within a few percent while cutting simulation time. The authors stress that future progress depends on explainable, physics-guided models and on methods that work with weak or absent labels, concluding that the state of the art already shows 'very promising results' in diagnosing noise, predicting failures, and simulating components.

Core claim

The central claim is that the current state of the art already demonstrates 'very promising results' in three tasks: early detection of failures, diagnosis of noise sources, and flexible simulation of instrument components—tasks that were previously tedious or impossible to do by hand. The chapter grounds this in two end-to-end pipelines built for LIGO: one factorizes a channels-by-events-by-features tensor to select a handful of diagnostic channels that act as veto witnesses for glitches, validated by true/false-positive rate curves; the other applies Isolation Forest to multivariate segments preceding lock losses to flag outlier 'precursor' states without explicit labels. For design, the c

What carries the argument

The load-bearing tools are three. Tensor factorization (CP/PARAFAC) reduces a 3-mode tensor encoding coincident glitches in roughly 900 diagnostic channels into latent factors; the channel-mode factor matrix selects candidate witness channels. Isolation Forest, an unsupervised tree-based outlier detector, partitions data preceding control failures to isolate anomaly precursors without labels. And generative adversarial networks (GANs)—a generator/discriminator pair trained adversarially—serve as differentiable, fast stand-ins for physics simulators; the cited calorimeter example uses a physics-informed loss to generate three-dimensional shower energy depositions.

Load-bearing premise

The design-side claim relies on the assumption that a GAN trained on simulator output is a faithful enough stand-in that optimizing against it yields a design that works in the real instrument—an assumption the chapter itself flags when it notes that 'certain features of some distributions are not well described.'

What would settle it

Build a design campaign using a GAN-trained surrogate, then run the selected design through the full physics simulator (or a prototype) and compare predicted and actual evaluation metrics; a systematic discrepancy would refute the design-optimization claim. A lighter test: measure whether the GAN reproduces correlations among shower-shape variables, not just marginal statistics.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If correct, LIGO's unidentified 20–60 Hz noise can be attacked with unsupervised learning, potentially enlarging the observable universe volume and enabling detection of the most massive black-hole mergers.
  • Automatically selected witness channels can veto glitches in science data, reducing false triggers and improving the speed and reliability of real-time gravitational-wave alerts.
  • GAN-based simulation makes simulation runtime independent of particle energy, enabling rapid exploration of design spaces that were computationally prohibitive.
  • Modular neural simulators could eventually be chained into end-to-end instrument simulation, changing how future accelerators like the Electron-Ion Collider are designed.
  • The same weak-supervision and explainability methods would generalize to other complex engineered systems with abundant sensor data but scarce failure labels.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A testable consequence the chapter leaves implicit: a design found by optimizing against a GAN surrogate should be verified by re-running the original simulator; the divergence between predicted and actual performance is a measure of the surrogate's fidelity.
  • The witness-channel and precursor pipelines could transfer to other sensor-rich facilities—fusion devices, telescopes, power grids—where failure labeling is costly, since both are unsupervised.
  • The chapter's emphasis on physics-guided GANs suggests a path toward 'calibrated surrogates' in which the model is constrained by conservation laws, potentially reducing the distributional mismatch it acknowledges.
  • If GANs become reliable simulators, the bottleneck in instrument design shifts from computation to the quality and coverage of training data, raising questions about how to validate diversity and physicality of generated data.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. This review chapter argues that machine learning can improve both the operation and the design of large scientific instruments. After laying out general challenges (weak supervision, explainability, physics-guided modeling, human-in-the-loop), it presents two LIGO case studies: tensor factorization to identify witness channels for noise transients, and Isolation Forest to detect precursors to lock-loss control failures. For instrument design, it reviews traditional simulation/optimization and then discusses GAN-based surrogates for physics simulation, focusing on CaloGAN for calorimeter showers. The chapter concludes that ML shows 'very promising results' for early failure detection, noise-source diagnosis, and flexible component simulation. The paper is written as a book chapter and relies on published work rather than presenting new experiments.

Significance. The chapter is a serviceable and readable synthesis of an important emerging area. Its strengths include the concrete grounding in LIGO operational problems, the clear articulation of the unsupervised-learning challenges specific to instruments, and pointers to public datasets. The operation-diagnosis claims are broadly consistent with the cited literature. The design-side claim, however, is weaker than the conclusion suggests: the only worked example is CaloGAN, whose own validation text (reproduced in the chapter) acknowledges distributional mismatch and cross-domain classifier degradation. Because no test connects GAN fidelity to any design decision, the statement that ML is 'enabling' design should be treated as a research promise, not a demonstrated result. This distinction is important for the chapter's central message.

major comments (2)
  1. [§7.5.2, §7.6] The design half of the conclusion ('enabling the flexible simulation of different components of an instrument') rests on CaloGAN as a substitute for GEANT4. The chapter itself reproduces the CaloGAN paper's caveat that 'certain features of some distributions are not well described' and that 'Training on CaloGAN and testing on GEANT4 does show significant degradation.' No end-to-end test is provided that an optimization objective computed on GAN-generated data ranks designs correctly relative to the true simulator. Thus, 'enabling ... design' is an extrapolation from distribution-level plausibility checks. Please either soften §7.6 to say 'potentially enabling' with an explicit statement that design-optimization fidelity remains to be demonstrated, or add a case where a design quantity of interest is validated on GAN output against simulator output.
  2. [§7.4.1–7.4.2] The two LIGO examples are presented as 'detailed real-world examples,' but the reported support is qualitative. For the tensor-factorization witness selection, the text refers to true- and false-positive rates but gives no values; Fig. 7.6 is a schematic. For the Isolation Forest lock-loss analysis, Fig. 7.7 is a t-SNE projection with no quantitative outlier-detection metrics (e.g., precision/recall, ROC/AUC, or comparison to a baseline). Since these pipelines are the main evidence for the operational-performance claim, please add the actual performance numbers or state explicitly that the results are preliminary and point to [19], [22] for quantitative details.
minor comments (6)
  1. [§7.2] Typographical errors: 'through review' should be 'thorough review', and 'indispensible' should be 'indispensable'.
  2. [§7.4] 'neutron start merger' should be 'neutron star merger'.
  3. [§7.4] The cross-references 'Section 1.3.1 and 1.3.2' should be §7.4.1 and §7.4.2.
  4. [Fig. 7.8] The caption contains a large verbatim block from [37] (the CaloGAN PRL paper), apparently inserted into the figure caption. If intended as a quotation, it should be shortened, clearly attributed, and typeset as a quotation rather than a figure caption.
  5. [Fig. 7.2] There are encoding artifacts in the rendered text (e.g., '□23', '10□22'); ensure the final typesetting renders the axes and math correctly.
  6. [References] Reference [22] is an internal LIGO technical report, and [19] is a workshop contribution. Please flag these as non-peer-reviewed sources when they are used to support the main operational examples.

Circularity Check

0 steps flagged

No significant circularity: the chapter is a review that cites external empirical work; self-citations are evidence, not redefinitions.

full rationale

This is a survey/review chapter, not a derivation chain. The strongest operational claims (tensor-factorization glitch witness selection and Isolation-Forest lock-loss precursors) are presented as summaries of previously published work: the glitch pipeline is explicitly attributed to 'In [19], we use matrix and tensor factorization...' and the lock-loss statistics to '[22]'. Although [19] shares authors with the chapter, it is an external empirical pipeline with its own validation figures (true/false positive rates), so citing it is not circular in the load-bearing sense defined by the review rules. No parameter is fitted to a subset of data and then renamed as a prediction; no definition depends on the conclusion. The design half relies on CaloGAN [37], an external paper, and the chapter honestly reproduces that paper's limitation: 'certain features of some distributions are not well described' and 'Training on CaloGAN and testing on GEANT4 does show significant degradation.' That admission undercuts the strength of the 'enabling... simulation for design' conclusion as a correctness/evidence matter, but it does not make the conclusion equivalent to its input or reduce by construction. The chapter contains no 'uniqueness theorem' or ansatz smuggled in via self-citation, and no known result is renamed as new. Accordingly, the appropriate finding is no significant circularity; the GAN-fidelity caveat is a correctness risk, not a circularity finding.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The paper does not introduce new entities or fitted parameters; however, the described ML pipelines depend on unstated hyperparameters and on domain assumptions about glitch coincidence and failure precursors. The GAN design application assumes generated data is a faithful surrogate, which the cited CaloGAN work itself qualifies.

free parameters (3)
  • Number of tensor factorization components N
    Section 7.4.1: 'We factorize this tensor into N factors' — N is a hyperparameter chosen by hand; the example figure uses N=5 (Figure 7.5 caption) but no selection criterion is given.
  • Isolation Forest parameters (contamination, tree count)
    Section 7.4.2: the outlier detection uses an Isolation Forest but no parameter values are stated; the outlier-score threshold for declaring precursors is also unspecified.
  • Coincidence window and feature set for glitch tensor
    Section 7.4.1: 'coincident ... within a short window' — window duration and glitch features (duration, peak frequency, etc.) are not specified.
axioms (4)
  • domain assumption Coincidence between a diagnostic-channel glitch and a main-channel glitch within a short time window indicates a potential causal witness relationship.
    Section 7.4.1, Figure 7.3 middle row: the pipeline relies on this to select witness channels.
  • domain assumption Data segments preceding control failures contain anomalies that are precursors to the failure.
    Section 7.4.2: 'We hypothesize that data preceding failure events in a subset of channels will have anomalies which might be precursors to the failures.'
  • domain assumption A GAN that reproduces the training distribution of simulator outputs can substitute for the simulator in design optimization.
    Section 7.5.2: the ML-based simulation pipeline replaces the simulator with a GAN generator; the paper acknowledges generated data may not match all statistical features.
  • domain assumption The unidentified 20–60 Hz excess noise in LIGO is non-linear/non-Gaussian and not coherent with auxiliary channels.
    Section 7.2, Figure 7.2: this motivates the unsupervised approach; it is cited from LIGO noise budget literature.

pith-pipeline@v1.3.0-alltime-deepseek · 18770 in / 13190 out tokens · 129666 ms · 2026-08-02T01:32:55.730013+00:00 · methodology

0 comments
read the original abstract

In modern experimental physics, particle accelerators and gravitational-wave observatories enable a wide-range of research at the frontiers of science. These instruments are highly complex consisting of many interacting systems which can face significant operational challenges. Apart from the experiment's main data product, a lot of data about the experimental apparatus and its environment is recorded. Machine learning techniques can analyze this big data at scale and find useful insights into operational faults potentially improving the instrument's performance and achieving the design goals. Speaking of design, machine learning can also accelerate/augment the expensive physics simulations used during the design phase of such large-scale instruments.

Figures

Figures reproduced from arXiv: 2607.14619 by Barry C. Barish, Evangelos E. Papalexakis, Jonathan W. Richardson, Rutuja Gurav.

Figure 7
Figure 7. Figure 7: Inside the control room of the Laser Interferometer Gravitational Observatory [PITH_FULL_IMAGE:figures/full_fig_p001_7.png] view at source ↗
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p003_7.png] view at source ↗
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 7
Figure 7. Figure 7: Time-frequency representation of the gravitational-wave event GW170817, as [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 1
Figure 1. Figure 1: FIG. 1. Average γ GEANT CaloGAN shower (bottom r generated by a simulator [PITH_FULL_IMAGE:figures/full_fig_p021_1.png] view at source ↗
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p021_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

60 extracted references · 3 canonical work pages

  1. [1]

    doi:10.1088/0264-9381/31/24/245010 , url =

    A Staley and D Martynov and R Abbott and R X Adhikari and K Arai and S Ballmer and L Barsotti and A F Brooks and R T DeRosa and S Dwyer and A Effler and M Evans and P Fritschel and V V Frolov and C Gray and C J Guido and R Gustafson and M Heintze and D Hoak and K Izumi and K Kawabe and E J King and J S Kissel and K Kokeyama and M Landry and D E McClelland...

  2. [2]

    arXiv preprint arXiv:2109.02551 , year=

    Pith/arXiv arXiv

  3. [3]

    Classical and quantum gravity , volume=

    Gravity Spy: integrating advanced LIGO detector characterization, machine learning, and citizen science , author=. Classical and quantum gravity , volume=. 2017 , publisher=

    2017

  4. [4]

    Journal of Machine Learning Research , year =

    Laurens van der Maaten and Geoffrey Hinton , title =. Journal of Machine Learning Research , year =

  5. [5]

    SIAM review , volume=

    Tensor decompositions and applications , author=. SIAM review , volume=. 2009 , publisher=

    2009

  6. [6]

    Advances in Neural Information Processing Systems 30 , editor =

    A Unified Approach to Interpreting Model Predictions , author =. Advances in Neural Information Processing Systems 30 , editor =. 2017 , publisher =

    2017

  7. [7]

    2008 eighth ieee international conference on data mining , pages=

    Isolation forest , author=. 2008 eighth ieee international conference on data mining , pages=. 2008 , organization=

    2008

  8. [8]

    SoftwareX , volume=

    Omicron: a tool to characterize transient noise in gravitational-wave detectors , author=. SoftwareX , volume=. 2020 , publisher=

    2020

  9. [9]

    AAAI 2020 Fall Symposium on Physics-Guided AI to Accelerate Scientific Discovery , year=

    Unsupervised matrix and tensor factorization for LIGO glitch identification using auxiliary channels , author=. AAAI 2020 Fall Symposium on Physics-Guided AI to Accelerate Scientific Discovery , year=

    2020

  10. [10]

    Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=

    Predicting particle accelerator failures using binary classifiers , author=. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=. 2020 , publisher=

    2020

  11. [11]

    PloS one , volume=

    On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation , author=. PloS one , volume=. 2015 , publisher=

    2015

  12. [12]

    On-line Optimization of European XFEL with OCELOT , author=. Proc. 16th Int. Conf. on Accelerator and Large Experimental Control Systems (ICALEPCS’17) , pages=

  13. [13]

    Proceedings of the European Conference on Computer Vision (ECCV) , pages=

    How good is my GAN? , author=. Proceedings of the European Conference on Computer Vision (ECCV) , pages=

  14. [14]

    Physical review letters , volume=

    Accelerating science with generative adversarial networks: an application to 3D particle showers in multilayer calorimeters , author=. Physical review letters , volume=. 2018 , publisher=

    2018

  15. [15]

    Fast simulation of the ATLAS calorimeter system with Generative Adversarial Networks , author=

  16. [16]

    Computing and Software for Big Science , volume=

    Learning particle physics by example: location-aware generative adversarial networks for physics synthesis , author=. Computing and Software for Big Science , volume=. 2017 , publisher=

    2017

  17. [17]

    npj Computational Materials , volume=

    Generative adversarial networks (GAN) based efficient sampling of chemical composition space for inverse design of inorganic materials , author=. npj Computational Materials , volume=. 2020 , publisher=

    2020

  18. [18]

    Journal of Cheminformatics , volume=

    Mol-CycleGAN: a generative model for molecular optimization , author=. Journal of Cheminformatics , volume=. 2020 , publisher=

    2020

  19. [19]

    Physics Letters B , volume=

    Deep Learning for real-time gravitational wave detection and parameter estimation: Results with Advanced LIGO data , author=. Physics Letters B , volume=. 2018 , publisher=

    2018

  20. [20]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    A style-based generator architecture for generative adversarial networks , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  21. [21]

    Communications of the ACM , volume=

    Generative adversarial networks , author=. Communications of the ACM , volume=. 2020 , publisher=

    2020

  22. [22]

    arXiv preprint arXiv:2104.10201 , year=

    Bayesian optimization is superior to random search for machine learning hyperparameter tuning: Analysis of the black-box optimization challenge 2020 , author=. arXiv preprint arXiv:2104.10201 , year=

    Pith/arXiv arXiv 2020

  23. [23]

    Proceedings of the 7th International Particle Accelerator Conference , year=

    Bayesian optimization of FEL performance at LCLS , author=. Proceedings of the 7th International Particle Accelerator Conference , year=

  24. [24]

    arXiv preprint arXiv:1306.2973 , year=

    Finesse, frequency domain INterferomEter simulation softwarE , author=. arXiv preprint arXiv:1306.2973 , year=

    Pith/arXiv arXiv

  25. [25]

    Nuclear instruments and methods in physics research section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=

    GEANT4—a simulation toolkit , author=. Nuclear instruments and methods in physics research section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=. 2003 , publisher=

    2003

  26. [26]

    Journal of intelligent information systems , volume=

    The human is the loop: new directions for visual analytics , author=. Journal of intelligent information systems , volume=. 2014 , publisher=

    2014

  27. [27]

    , author=

    Name Disambiguation in AMiner: Clustering, Maintenance, and Human in the Loop. , author=. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining , pages=

  28. [28]

    Proceedings of the second workshop on data management for end-to-end machine learning , pages=

    Accelerating human-in-the-loop machine learning: Challenges and opportunities , author=. Proceedings of the second workshop on data management for end-to-end machine learning , pages=

  29. [29]

    arXiv e-prints , pages=

    Active Anomaly Detection for time-domain discoveries , author=. arXiv e-prints , pages=

  30. [30]

    2009 , publisher=

    Active learning literature survey , author=. 2009 , publisher=

    2009

  31. [31]

    Journal of the American society for information science , volume=

    Improving retrieval performance by relevance feedback , author=. Journal of the American society for information science , volume=. 1990 , publisher=

    1990

  32. [32]

    Advances in Neural Information Processing Systems , volume=

    Crowdsourced clustering: Querying edges vs triangles , author=. Advances in Neural Information Processing Systems , volume=

  33. [33]

    IEEE Transactions on knowledge and data engineering , volume=

    Theory-guided data science: A new paradigm for scientific discovery from data , author=. IEEE Transactions on knowledge and data engineering , volume=. 2017 , publisher=

    2017

  34. [34]

    ACM computing surveys (CSUR) , volume=

    A survey of methods for explaining black box models , author=. ACM computing surveys (CSUR) , volume=. 2018 , publisher=

    2018

  35. [35]

    Journal of Physics: Conference Series , volume=

    Anomaly detection for beam loss maps in the Large Hadron Collider , author=. Journal of Physics: Conference Series , volume=. 2017 , organization=

    2017

  36. [36]

    International Cross-Domain Conference for Machine Learning and Knowledge Extraction , pages=

    Explainable Deep Learning for Fault Prognostics in Complex Systems: A Particle Accelerator Use-Case , author=. International Cross-Domain Conference for Machine Learning and Knowledge Extraction , pages=. 2020 , organization=

    2020

  37. [37]

    Mechanical Systems and Signal Processing , volume=

    A review on the application of deep learning in system health management , author=. Mechanical Systems and Signal Processing , volume=. 2018 , publisher=

    2018

  38. [38]

    Matrix Profile XXIII: Contrast Profile: A Novel Time Series Primitive that Allows Real World Classification , url=

  39. [39]

    2017 IEEE international conference on data mining (ICDM) , pages=

    Matrix profile VI: Meaningful multidimensional motif discovery , author=. 2017 IEEE international conference on data mining (ICDM) , pages=. 2017 , organization=

    2017

  40. [40]

    Machine Learning: Science and Technology , volume=

    Enhancing gravitational-wave science with machine learning , author=. Machine Learning: Science and Technology , volume=. 2020 , publisher=

    2020

  41. [41]

    Philosophical Transactions of the Royal Society A , volume=

    Machine learning and big scientific data , author=. Philosophical Transactions of the Royal Society A , volume=. 2020 , publisher=

    2020

  42. [42]

    arXiv preprint arXiv:1811.03172 , year=

    Opportunities in machine learning for particle accelerators , author=. arXiv preprint arXiv:1811.03172 , year=

    Pith/arXiv arXiv

  43. [43]

    Unsupervised Machine Learning for Detection of Faulty Beam Position Monitors , author=. Proc. 10th Int. Particle Accelerator Conf. (IPAC’19), Melbourne, Australia , volume=

  44. [44]

    IEEE Transactions on Plasma Science , volume=

    Exploratory data analysis techniques to determine the dimensionality of complex nonlinear phenomena: The L-to-H transition at JET as a case study , author=. IEEE Transactions on Plasma Science , volume=. 2012 , publisher=

    2012

  45. [45]

    2012 16th IEEE Mediterranean Electrotechnical Conference , pages=

    Identifying jet instabilities with neural networks , author=. 2012 16th IEEE Mediterranean Electrotechnical Conference , pages=. 2012 , organization=

    2012

  46. [46]

    Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=

    On the identification of instabilities with neural networks on JET , author=. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=. 2013 , publisher=

    2013

  47. [47]

    Proceedings of the European Particle Accelerator Conference , pages=

    A beam diagnostic system for accelerator using neural networks , author=. Proceedings of the European Particle Accelerator Conference , pages=

  48. [48]

    particle acceleration control , author=

    Using archon-3. particle acceleration control , author=. IEEE Expert , volume=. 1996 , publisher=

    1996

  49. [49]

    1 , author=

    Using archon to develop real-world dai applications. 1 , author=. IEEE expert , volume=. 1996 , publisher=

    1996

  50. [50]

    Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=

    OCELOT: A software framework for synchrotron light source and FEL studies , author=. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , volume=. 2014 , publisher=

    2014

  51. [51]

    Brustein, Ram and Medved, A. J. M. , journal =. 2018 , month = feb, publisher =. doi:10.1103/PhysRevD.97.044035 , url =

    work page doi:10.1103/physrevd.97.044035 2018

  52. [52]

    Nature , keywords =

    A two per cent Hubble constant measurement from standard sirens within five years. Nature , keywords =. doi:10.1038/s41586-018-0606-0 , archivePrefix =. 1712.06531 , primaryClass =

    Pith/arXiv arXiv

  53. [53]

    and Baiotti, Luca and Creighton, Jolien D

    Read, Jocelyn S. and Baiotti, Luca and Creighton, Jolien D. E. and Friedman, John L. and Giacomazzo, Bruno and Kyutoku, Koutarou and Markakis, Charalampos and Rezzolla, Luciano and Shibata, Masaru and Taniguchi, Keisuke , journal =. 2013 , month = aug, publisher =. doi:10.1103/PhysRevD.88.044042 , url =

    work page doi:10.1103/physrevd.88.044042 2013

  54. [54]

    Environmental influences on the

    A Effler and R M S Schofield and V V Frolov and G Gonz. Environmental influences on the. Classical and Quantum Gravity , abstract =. doi:10.1088/0264-9381/32/3/035017 , url =

    doi:10.1088/0264-9381/32/3/035017

  55. [55]

    Environmental noise in

    P Nguyen and R M S Schofield and A Effler and C Austin and V Adya and M Ball and S Banagiri and K Banowetz and C Billman and C D Blair and A Buikema and C Cahillane and F Clara and P B Covas and G Dalya and C Daniel and B Dawes and R DeRosa and S E Dwyer and R Frey and V V Frolov and D Ghirado and E Goetz and T Hardwick and A F Helmling-Cornell and I J Ho...

  56. [56]

    L Blackburn and L Cadonati and S Caride and S Caudill and S Chatterji and N Christensen and J Dalrymple and S Desai and A Di Credico and G Ely and J Garofoli and L Goggin and G Gonz. The. Classical and Quantum Gravity , abstract =. doi:10.1088/0264-9381/25/18/184004 , url =

    work page doi:10.1088/0264-9381/25/18/184004

  57. [57]

    Classical and Quantum Gravity , abstract =

    M Cabero and A Lundgren and A H Nitz and T Dent and D Barker and E Goetz and J S Kissel and L K Nuttall and P Schale and R Schofield and D Davis , title =. Classical and Quantum Gravity , abstract =. doi:10.1088/1361-6382/ab2e14 , url =

    doi:10.1088/1361-6382/ab2e14

  58. [58]

    Finesse , Url =

    Brown, Daniel David and Freise, Andreas , Doi =. Finesse , Url =

  59. [59]

    Brown and Philip Jones and Samuel Rowlinson and Sean Leavey and Anna C

    Daniel D. Brown and Philip Jones and Samuel Rowlinson and Sean Leavey and Anna C. Green and Daniel Töyrä and Andreas Freise , keywords =. Pykat: Python package for modelling precision optical interferometers , journal =. 2020 , issn =. doi:https://doi.org/10.1016/j.softx.2020.100613 , url =

    arXiv 2020

  60. [60]

    Hiro Yamamoto , Title =